Honing test methods to detect device-related BSIs
Choice of weapons against a formidable infection
Synopsis: In a meta-analysis of studies assessing methodologies for the diagnosis of intravascular device-related bloodstream infections, paired quantitative blood cultures drawn from the catheter and a peripheral site are the most accurate. However, numerous other methods, including quantitative catheter culture, semiquantitative catheter culture, and differential time to blood culture positivity have sufficient sensitivity and specificity to be clinically useful.
Source: Safdar N, et al. Meta-analysis: Methods for diagnosing intravascular device-related bloodstream infection. Ann Intern Med 2005; 142:451-466.
Intravascular device-related bloodstream infection (IDR-BSI) are a major source of morbidity and mortality. An estimated 250,000 infections occur annually in the United States, with an attributable mortality of 12% to 25%.1
Accurate diagnosis is essential to management of IDR-BSI. A number of diagnostic methods have been proposed and evaluated in clinical studies, but there is no consensus as to the most accurate method of diagnosis.
Safdar and colleagues reviewed the English-language literature published between 1966 and 2004 for studies evaluating methods for diagnosis of IDR-BSI.
The 51 studies included in the meta-analysis compared the method being evaluated against a reference standard, and contained sufficient data to allow calculation of sensitivity and specificity. Safdar, et al studied eight diagnostic methods commonly used in clinical practice. (See chart.)
They calculated the overall sensitivity and pooled specificity of each method and determined summary measures of accuracy using receiver operating curves and log odds ratio. They also assessed test accuracy for long-term (tunneled, cuffed, or implanted) catheters and short-term catheters.
Because different studies used different reference standards for the evaluation of individual tests, Safdar, et al performed multiple subgroup analyses according to the type of reference standards used.
The most accurate tests were paired quantitative culture and quantitative blood culture through the device.
Of the methods requiring catheter removal, quantitative catheter segment culture was the most accurate.
However, the other tests studied had acceptable performance with sensitivity and specificity generally exceeding 75%.
Not surprisingly, all tests showed better performance with increasing prevalence of IDR-BSI, and all tests showed a negative predictive value of > 95% at a prevalence of 20%.
There was a significant difference in accuracy when certain tests were applied to short-term catheters compared with long-term catheters.
Semiquantitative catheter culture was more accurate when applied to short-term catheters; differential time to culture positivity was more accurate when applied to long-term catheters.
Based on their findings, Safdar, et al recommended that diagnostic testing for IDR-BSI be performed when the probability was significant (e.g., > 0.2).
For short-term catheters, they recommended quantitative or semiquantitative catheter segment culture combined with two blood cultures, with one drawn peripherally and one drawn through the catheter.
For long-term catheters, they recommended paired quantitative blood cultures as the most accurate method, but noted that differential time to positivity of paired blood cultures provided similar sensitivity with adequate specificity.
Comment by Robert Muder, MD, hospital epidemiologist, Pittsburgh VA Medical Center.
Given that IDR-BSI is so frequent and extracts such a heavy toll, it’s surprising that there is no consensus as to the best diagnostic approach. Safdar, et al. are to be commended for undertaking such an ambitious analysis and for providing the infectious disease community with credible estimates of the performance of individual tests that can be used to formulate rationale diagnostic strategies.
Their analysis has a number of limitations, which they point out:
The studies used different reference standards, and subgroup analyses revealed significant heterogeneity in test performance based on the reference standard used. The analysis did not take into account the potential effect of previous administration of antimicrobials.
However, the results appear sufficiently robust that those weaknesses would appear to be relatively minor. In practice, one issue that needs to be considered is the laboratory effort required to perform individual tests.
Quantitative tests, such as quantitative paired cultures and quantitative catheter segment culture, are more accurate than their semiquantitative or qualitative counterparts, but require considerably more time and effort to perform.
Thus, tests such as differential time to blood culture positivity and semiquantitative catheter culture are attractive since they provide acceptable accuracy with minimal impact on the laboratory workload.
- O’Grady NP, et al. Guidelines for the prevention of intravascular catheter-related infections. MMWR 2002; 51(RR10):1-29.